Remove NT Default

[Sligneris]

 

NT Default has a bunch of serious issues that I noticed quite early while playing AI.

 

While Baystation, the server that NT Default originates from, has no law priority, our server... well, quite obviously does. NT Default works fine on Baystation, but on a server where law priority is a thing, it has severe issues.

 

For example, protecting the station's integrity comes before everything else, including serving or protecting the crew. That means that if the crew damages the station, it's implied that the AI can use any means necessary to stop them - I'm not sure how far would the AI go in trying to resolve a law conflict like that.

 

Next, serving the crew comes before protecting, which means that the AI might believe it is allowed to kill on order - as long as it serves the crew or protects the station integrity. You know, the terrifying "a robot can kill a human if it serves humanity" thing from the Asimov books, except with "an AI can kill a crewmember if it serves the crew"? Having a clueless AI player exercise this kind of thinking, while believing they're following their laws can turn ugly.

 

Honestly, I'm not even sure how much these are intended - since all I've been told while mentioning these issues upon ahelping is "follow the spirit"... Which honestly doesn't much tell me how am I supposed to act as an AI under this. My best judgment tells me these were not meant to work this way, but then why have a lawset that conditions the AI to do things it really shouldn't?

 

When a plain interpretation can turn this bad, I'm starting to doubt the lawset itself. Especially with the problems regarding the first law, I think having AI follow this really can't end well. So, I recommend it's either removed (Crewsimov works just fine to serve an almost identical purpose), or at least has its order rearranged, depending on the priority it's really supposed to have.

 

Edited March 7, 2016 by Guest

[pinatacolada]

 

My interpretation of default is that I am here to serve the crew as a whole. Opening the bridge to the clown doesn't serve the crew, locking down a traitor with something dangerous, does. And if he's exceptionally violent and destructive, I might even bat an eye at a lynching. The whole protect the station doesn't allow me to kill anyone though, there's far better way to stop them usually. Will I ignore screams for help while I am trying to contain a breach? Yes. Tough luck.

 

Different people have slightly different interpretations of the laws, and that's the !!FUN!! of playing AI.

 

[Sligneris]

Honestly, the thing that concerns me the most is the implications of protecting the integrity of the station itself is before all else. Obviously, your way of handling this does seem to be the best, but there were a few points that I saw as red flags, sorta.

[Coldflame]

 

Protecting the 'station' generally includes the crew and the ability for the crew to function, given the station as a whole is intended to be manned and function.

 

It's entirely up to the AI, as with most lawsets. If someone wants to be a dick as an AI, you need a completely watertight lawset to stop them, which none of the current lawsets do anyway.

 

THAT SAID, I still think that a lot of the current lawsets should be removed as a default choice. I'd argue corporate as the default lawset every round, but that also has a fair share of flaws.

 

Being completely honest- I'm not sure this is actually a huge issue. Generally, things are fine as they are.

 

[TullyBBurnalot]

 

This really only becomes an issue when the AI player decides to legitimately ignore all context and just go with the literalist interpretation of the lawset.

 

Rules > Roleplay, at all times. All lawsets can be abused to hell and back if you go with the letter of it, but that doesn't mean you should do it.

 

[sdtwbaj]

 

This really only becomes an issue when the AI player decides to legitimately ignore all context and just go with the literalist interpretation of the lawset.

 

Rules > Roleplay, at all times. All lawsets can be abused to hell and back if you go with the letter of it, but that doesn't mean you should do it.

Honestly, a computer only knows what you'd tell it. They aren't pAI, they can't, or shouldn't, understand the spirit of their laws, just the literalist interpretation.

 

[TullyBBurnalot]

 

A literalist interpretation of, say, Crewsimov, would mean the AI would have to listen to anyone who threatens suicide unless the AI does something, a situation that is explicitly referenced in the Server Rules, where it states the AI is free to ignore it.

 

We do not expect AI players to behave like IRL robots because, quite frankly, that would lead to the AI players being given free reign to break a fuckton of Server Rules based entirely on a strict, literalist interpretation of their lawset. And even if no Server Rules were broken, a strict, literalist interpretation that completely ignores any and all context is not fun for anyone except people who abuse the situation.

 

Two points here.

 

First, Rules > Roleplay.

 

Secondly, the Golden Rule. Ask yourself "Is what I'm doing going to significantly impact on the rest of the player's enjoyment of the round?". If yes, then don't do it.

 

I've said it before, and I'll keep saying it. In the context of this game, literalist interpretations of lawsets that completely ignore context and circumstances are bad.

 

[necaladun]

This hasn't come up as a problem that I've seen. Idiot-proofing laws is useless, the internet will always breed a better idiot. The solution is not to be an idiot, and if someone is, they get a jobban.

[Earthdivine]

 

This hasn't come up as a problem that I've seen. Idiot-proofing laws is useless, the internet will always breed a better idiot. The solution is not to be an idiot, and if someone is, they get a jobban.

Well I mean, that's the reason we got rid of round-start NT Default, because the AI's were being shitlers and killing anyone that harms the station or remotely looked antaggy.